latexresume
Open-source LaTeX resume builder with AI assistant and MCP server
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/adhi2624--latexresume)HTML
<a href="/listings/adhi2624--latexresume"><img src="/api/listings/adhi2624--latexresume/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Requirements: Node 20.9+, [pnpm](https://pnpm.io) 9 (`corepack enable`), Docker.
pnpm install cp .env.example .env.local # then set JWT_SECRET and ENCRYPTION_SECRET (openssl rand -base64 48) pnpm services # MongoDB Community + LaTeX compiler (docker compose --env-file .env.local up -d) pnpm dev # http://localhost:3000
Already running MongoDB on 27017? Use `docker compose up -d compiler`, or change the host port with `MONGO_PORT=27018 docker compose up -d` and update `MONGODB_URI`.
Create an account at `/auth/signup`, pick a template, press **Compile**.
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 112 lines
- # LaTeXResume
- An open-source, self-hostable **LaTeX resume builder** with live PDF preview, an **AI assistant**
- (bring your own key) and a built-in **MCP server**. Run it locally with Docker, or deploy it to
- **Cloudflare Workers** with **MongoDB Atlas**.
- - ✍️ Monaco editor + real TeX Live compilation (your resume is just a `.tex` file)
- - 🤖 AI chat with OpenAI, Anthropic or any OpenAI-compatible endpoint (Ollama, OpenRouter, vLLM…); keys are stored AES-GCM encrypted
- - 🔌 MCP server at `/mcp` — let Claude, Cursor, etc. list, edit, compile and publish resumes
- - 🔗 Publish a resume to a public URL (and unpublish)
- - 🪶 No plans, billing or tracking. MIT licensed.
- ## Architecture
- ```
- Browser ──► Next.js app (Node, or Cloudflare Worker via OpenNext) ──► MongoDB (Community locally / Atlas in prod)
- │
- └──► LaTeX compiler service (Docker, TeX Live) ← Workers cannot run TeX
- ```
- Workers can't execute `pdflatex`, so compilation lives in a small separate service (`compiler/`).
- Run it with Docker anywhere (Fly.io, Railway, a VPS, Cloud Run, Cloudflare Containers…) and point
- `LATEX_COMPILER_URL` at it.
- ## Quick start (local)
- Requirements: Node 20.9+, [pnpm](https://pnpm.io) 9 (`corepack enable`), Docker.
- ```bash
- pnpm install
- cp .env.example .env.local # then set JWT_SECRET and ENCRYPTION_SECRET (openssl rand -base64 48)
- pnpm services # MongoDB Community + LaTeX compiler (docker compose --env-file .env.local up -d)
- pnpm dev # http://localhost:3000
- ```
- Already running MongoDB on 27017? Use `docker compose up -d compiler`, or change the host port with
- `MONGO_PORT=27018 docker compose up -d` and update `MONGODB_URI`.
- Create an account at `/auth/signup`, pick a template, press **Compile**.
- ## Configuration
- | Variable | Required | Description |
- | --- | --- | --- |
- | `MONGODB_URI` | ✅ | Connection string; the database name comes from the path (`…/latexresume`). |
- | `JWT_SECRET` | ✅ | ≥16 chars; signs session cookies. `openssl rand -base64 48` |
- | `ENCRYPTION_SECRET` | ✅ | Encrypts users' stored AI API keys. |
- | `LATEX_COMPILER_URL` | ✅ | URL of the compiler service. |
- | `LATEX_COMPILER_TOKEN` | – | Shared secret (the compiler's `AUTH_TOKEN`). |
- | `ALLOW_SIGNUP` | – | `false` closes registration (recommended for personal instances). |
- | `OPENAI_API_KEY` | – | Server-wide default AI key; users may bring their own. |
- | `NEXT_PUBLIC_REPO_URL` | – | Shows GitHub links in the UI (**build time** — set it when running `pnpm build`/`pnpm deploy`). |
- | `ALLOW_PRIVATE_AI_URLS` | – | Production only: allow loopback/private AI base URLs (blocked by default to prevent SSRF). |
- | `RATE_LIMIT_SCALE` | – | Multiplier for built-in rate limits (default `1` in production, `100` in development). |
- | `TRUST_PROXY` | – | Set `true` only behind a reverse proxy that sets `X-Forwarded-For`; otherwise per-IP limits share one bucket. |
- | `ALLOW_CUSTOM_AI_URLS` | – | `false` disables user-supplied AI base URLs entirely (recommended on shared hosts). |
- Compiler service (`compiler/`) variables: `PORT` (9000), `AUTH_TOKEN`, `COMPILE_TIMEOUT_MS` (30000), `MAX_CONCURRENT` (2).
- **Never expose the compiler publicly without `AUTH_TOKEN`** — it executes TeX on behalf of callers.
- ## Deploy
- - **Vercel (app) + Render (compiler) + MongoDB Atlas:** see **[docs/DEPLOY-VERCEL-RENDER.md](docs/DEPLOY-VERCEL-RENDER.md)**.
- - **Cloudflare Workers + MongoDB Atlas:** below.
- ### Cloudflare Workers + MongoDB Atlas
- See **[docs/DEPLOY-CLOUDFLARE.md](docs/DEPLOY-CLOUDFLARE.md)**. In short:
- ```bash
- pnpm db:init # once, creates indexes (against Atlas)
- npx wrangler secret put MONGODB_URI # standard (non-SRV) Atlas string!
- npx wrangler secret put JWT_SECRET
- npx wrangler secret put ENCRYPTION_SECRET
- npx wrangler secret put LATEX_COMPILER_URL
- pnpm deploy
- ```
- ## MCP server
- Create an API token in **Settings → API tokens & MCP**, then add the server to your client:
- ```json
- { "mcpServers": { "latexresume": { "type": "http", "url": "https://YOUR-HOST/mcp",
- "headers": { "Authorization": "Bearer lrp_…" } } } }
- ```
- Tools: `list_templates`, `list_resumes`, `get_resume`, `create_resume`, `update_resume`,
- `delete_resume`, `compile_resume`, `publish_resume`, `unpublish_resume`. Details in [docs/MCP.md](docs/MCP.md).
- ## Development
- ```bash
- pnpm lint # typecheck
- pnpm test # end-to-end API tests (needs the app, MongoDB and compiler running)
- pnpm build # production build
- pnpm preview # build + run in the local Workers runtime
- ```
- ## Security notes
- - LaTeX is code. The compiler runs `pdflatex` with `-no-shell-escape`, paranoid file access, a timeout, and
- should run in an unprivileged container (the provided compose file drops capabilities and uses a read-only root).
- Don't expose it publicly without `AUTH_TOKEN`.
- - Passwords use PBKDF2-SHA256; sessions are signed `httpOnly` cookies that are revoked on logout; API tokens are stored as SHA-256 hashes.
- - Rate limits (MongoDB-backed): login 8 failures / 10 min per IP+email and 20 / 15 min per account, signup 10 / hour per IP, compile 30 / min per user with one in-flight compile per user. Behind a reverse proxy set `TRUST_PROXY=true`.
- - The AI base URL cannot target private/loopback hosts in production.
- - Report vulnerabilities as described in [SECURITY.md](SECURITY.md).
- ## Contributing
- PRs welcome — see [CONTRIBUTING.md](CONTRIBUTING.md). Licensed under the [MIT License](LICENSE).
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.